基于小波分析的人脸图像特征提取

Face Feature Location Using Wavelet Analysis

人脸图像特征提取是人脸识别的基础,采用小波分析技术,提出了一种快速的、具有良好鲁棒性的人脸图像特征提取方法。应用小波分析良好的时频局部化特性和多分辨率特性,选用合适的小波基对人脸图像进行小波变换,得到体现人脸特征的小波高频细节;经过水平和垂直投影运算,定位出眼睛对、鼻子、嘴唇和双颊等重要的人脸图像特征;再结合人脸的FAP参数集,确定出人脸识别所需要的人脸关键特征。实验结果表明,该方法具有较好的实时性和鲁棒性。

Face feature extraction is the foundation for face recognition and face feature location is the first step. A robust and fast face feature location algorithm based on the wavelet analysis technology are introduced. Firstly ,decompose the face image by utilizing wavelet analysis to decrease its dimension and get its details which contain the key features, Then it＇s easy to locate the features by the horizontal and vertical projective operation. At last,confirm the required key features by the FAP sets. Experimental results show that this algorithm is mostly quick and robust.